Raspberry Pi crashing when transferring large files via SMB

  • Hi,


    Apologies for bothering. I've been running OMV on a Raspberry Pi 3B (not the +), using a wired connection, and an external USB drive formatted as ext4. OMV is installed in the SD card.


    Everything has been working spectacularly well, but yesterday I ran into an issue when trying to copy some "large" files (15GB) over Samba from Mac to it. It happened at first and I thought it had been a one-off error. The pi froze, SMB dropped, couldn't SSH into it nor access the GUI. I thought I would have to unplug it, but after some minutes it resumed activity as normal, so I let it go. SSH'd into it and rebooted it just in case.


    The morning after tried again, same thing happened, and moments ago again so I'm thinking it's a fairly repeatable issue.


    Here's part of the syslog:


    https://pastebin.com/kCrL9uY4


    I ran into a few similar issues on this and other forums, but they seem to suggest that this had been fixed in a previous kernel update. I'm on


    Linux openmediavault 4.14.98-v7+ #1200 SMP Tue Feb 12 20:27:48 GMT 2019 armv7l GNU/Linux


    which should be good, given the last reports were from mid-last-year.


    https://forum.openmediavault.o…ng-up-gui-and-ssh-access/
    https://www.raspberrypi.org/fo…d24f1d39ae45d4bc&start=25
    https://www.raspberrypi.org/fo…208821&start=375#p1330384


    The main thing from the log is that it seems that smbd-notifyd eats up all the memory and then only when it gets killed does everything resume normal activity.


    I'm running "top" right now and watching things as they go, and indeed it's been steadily increasing and it's right now at 86% memory consumption, and starting to eat at the swap memory as well. EDIT: and it just died, after 9.28GB transferred, with 0 swap memory free and about 5MB free from the main memory :)


    Is there anything I'm missing or should be reporting on?


    Happy to help with anything. I imagine that I'll end up transferring this via rsync (maybe it'll happen the same) but wanted to share in case it helps - or if anyone has any more pointers.


    Thanks.

  • Hi @tkaiser, thanks for the reply - and thanks for all the work you do for the project!


    I believe it'll be http://ix.io/1F6G ?


    I got this in the prompt:


    Network/firewall problem detected. Not able to upload debug info.
    Please fix this or use "-U" instead and upload whole output manually


    System diagnosis information will now be uploaded to /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    /usr/bin/armbianmonitor: line 341: /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq: Permission denied
    http://ix.io/1F6G
    Please post the URL in the forum where you've been asked for.



    so then I ran it with sudo and -U and this is the output:


    https://pastebin.com/YVx17U1u


    In case it's different.


    Let me know if there's anything else I can help with!


    Edit: and just in case, running -u with sudo returns this: http://ix.io/1F6H

  • Thanks,
    (most probably) unrelated to the symptoms you're running into... but it seems your installation never finished the first boot setup. Can you please provide output from dpkg -l | grep raspberry?

    Sure!


    iilibraspberrypi-bin1.20190215-1 armhfMiscellaneous Raspberry Pi utilities
    iilibraspberrypi0 1.20190215-1 armhfEGL/GLES/OpenVG/etc. libraries for the Raspberry Pi's VideoCore IV
    iiraspberrypi-bootloader1.20190215-1 armhfRaspberry Pi bootloader
    iiraspberrypi-kernel1.20190215-1 armhfRaspberry Pi bootloader


    I could certainly use a different device if warranted, though I was kind of hoping to put this spare Raspberry Pi I had here to good use.


    I see quite a few comments around "Turning off tcp-segmentation offloading", though I struggle to make it run on the current OMV setup - doesn't seem to know about "ethtool", so I'm not especially keen on installing it if we're not using it so far.


    https://www.raspberrypi.org/fo…171122&p=1333422#p1333511


    Thoughts on how I could try the tcp-segmentation offloading approach before giving up on it?


    Thanks!

  • I struggle to make it run on the current OMV setup - doesn't seem to know about "ethtool"

    It's there but usually not in your $PATH. You can become superuser by doing a sudo -s and can then execute ethtool directly, otherwise you need to use the full path (and sudo too of course): sudo /sbin/ethtool


    Good luck with your RPi. I mean if you're really dealing with super large files why do you choose the slowest SBC on earth for this use case? The good/real ARM SBC transfer data at least 8 times faster. And I thought the offloading babbling over in the RPi forum only applies to their fancy 'new' RPi 3B+ with its crippled Gigabit Ethernet implementation?

  • Thanks - I'll give it a shot. It might certainly be related to the RPi 3B+ only - I didn't consider that. :)


    It's a fair point regarding the choice of device. I suppose in my case it's more a case of "I had it around and thought I'd use it". It's really a home server for my media, and not more - no instances of multiple users, really. I don't usually have such large files, it was actually a one off I had just put together. Your comment did give me pause, though, in terms of whether this would be an issue with reading as well, but I tested it moments ago and it doesn't seem the case.


    I'll also check if this happens only when a single file is over a certain size, or if it is for a single "copy" operation (i.e. multiple small files that add up to a large size) and see how to manage.


    I appreciate you looking into this, and will consider a different device if this becomes a more meaningful issue.


    Thank you and keep up the great work!

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!